What you have done by setting the car on a track and running it in a train does not give you a reliable test result. There are many variables, such as the resistance that is created by wheels, axles, rollers, track joints, etc.
Whenever you are confronted with an electrical problem, it's best to assemble the right tools to do the diagnosis and the repair. Do you own a digital volt-ohm meter? The cheaper ones that are suitable for toy trains cost from zero dollars (give-aways at Northern Tools) to around $12.00
Here's one for 12 bucks, with free shipping.
http://www.ebay.com/itm/231331359704
Do you have short flexible test leads that terminate in alligator clips? You should have a few on hand for doing tests.
I would put the car on its back, and carefully connect two test leads from a test transformer, or from one set of output binding posts on your ZW. Put one test lead directly on the frame of the car, and the other directly on the wire that runs from one roller assembly to the inside of the car. (NOT the roller, NOT the roller assembly, but the roller wire) Then, with the VM on the 0-20 Vac scale, start putting power on the car, using the lowest setting. Then increase the voltage until the car operates correctly. Decrease the voltage to a point where the car just stops working correctly. Make a note of the voltages. Then let us know what your findings are and get back to us for the next step in the process.
There is a possibility (or rather a probability) that the manufacturer tried to lower the cost by not providing an adequate gauge of wire, or some other short-cut. The lamps may be creating a voltage drop in the car. We will test that theory after we hear from you about the voltage range within which the car operates correctly in the test mode.